keystone keeps a lot of CPU during nova rally scenario

Bug #1469806 reported by Leontii Istomin
20
This bug affects 2 people
Affects Status Importance Assigned to Milestone
Mirantis OpenStack
Invalid
Medium
MOS Keystone

Bug Description

We have found that keystone keeps up to 900% CPU during boot_and_list_server_batch_x10 rally scenario.
atop from node-199: http://paste.openstack.org/show/325031/
atop from node-206 http://paste.openstack.org/show/325032/
atop from node-225 http://paste.openstack.org/show/325034/
We seen CPU load during ~3 minutes. At the time nova couldn't answer requests (500 HTTP)
http://paste.openstack.org/show/324755/
We can't reproduce the issue anymore.
Diagnostic Snapshot (needed logs have been deleted by logrotate, but atop binarys are there): http://mos-scale-share.mirantis.com/fuel-snapshot-2015-06-29_11-25-54.tar.xz

Tags: scale
Changed in mos:
status: New → Incomplete
status: Incomplete → Invalid
importance: Undecided → High
importance: High → Medium
Revision history for this message
Dina Belova (dbelova) wrote :

One more similar issue happened before - https://bugs.launchpad.net/mos/+bug/1467594 - I marked the old one as a duplicate for this one, probably more details can be found in https://bugs.launchpad.net/mos/+bug/1467594

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.